Abstract

Local image descriptors are generally designed for describ- ing all possible image patches. Such patches may be sub ject to complex variations in appearance due to incidental ob ject, scene and recording conditions. Because of this, a single-best descriptor for accurate image representation under all conditions does not exist. Therefore, we pro- pose to automatically select from a pool of descriptors the one that is best suitable based on ob ject surface and scene properties. These prop- erties are measured on the fly from a single image patch through a set of attributes. Attributes are input to a classifier which selects the best descriptor. Our experiments on a large dataset of colored ob ject patches show that the proposed selection method outperforms the best single descriptor and a-priori combinations of the descriptor pool.